特征驱动下区实体匹配指标自适应融合技术

Adaptive fusion technology of match indicators on area entity driven by data characteristics

空间实体匹配过程中多个指标的融合问题是影响匹配效果的关键问题之一。本文针对这一问题,以区实体为例提出了一套基于范例库的解决方案。首先提取出影响实体匹配的数据特征因子并确定了量化方法,其次选取典型的匹配指标,接下来通过建立指标权值范例库确定各指标权值,最后根据权值和数据特征因子调整匹配过程。该方法使得数据具有学习能力,达到了指标权值的自适应性的目标。实验表明该方法可行,并且可以提升空间实体匹配算法的效率、准确度和智能化程度。

The fusion of multiple indicators in spatial entities matching process is one of the key issues that affects the match effects. A solution based on sample libraries about area entities was proposed in the paper. First, the data match-related characteristics were extracted and the quantitative methods were determined. Second, typical match indicators were selected. Next, through the establishment of sample library, the weights of each indicator were determined. Finally according to the characteristics factors and weights the matching process was adjusted. This solution makes could data with ability to learn, to achieve the goal of indicator weights selfadaptivity. Experiments showed that the method is feasible and could improve the efficiency of spatial entities matching algorithms with high, accuracy and intelligence.